1. INTRODUCTION

The first release of the Gaia mission data (Gaia DR1) includes two astrometric catalogs (^{Brown and Gaia Collaboration 2016}). The smaller catalog, called TGAS, includes 2 million brighter stars with accurate proper motions and parallaxes and is based on a combination of astrometric data from Hipparcos and Tycho-2 (^{ESA 1997}; ^{Høg et al. 2000}) and Gaia itself (^{Lindegren et al. 2016}), while the larger catalog of 1.1 billion objects is derived from Gaia’s own observations and ICRF-2 radio source positions. I am using the TGAS in this paper, specifically, the parallaxes of brighter stars listed there. The formal errors of parallaxes are all smaller than 1 mas, which was the only requirement for an astrometric solution to be included in DR1. The entire set of 2 million Gaia DR1 stars was cross-matched with the GALEX DR5 catalogs by ^{Bianchi et al. (2011)}, namely, the All-Sky Imaging survey (AIS) with limiting magnitudes 19.9/20.8 in FUV/NUV and the Medium-depth Imaging Survey (MIS) with limiting magnitudes 22.6/22.7. The search for Galex matches was performed with Gaia J2015 positions in a cone of 3.5σ of Galex positions, but not greater than 5^′′on the sky. The total number of matched sources is 720622, which is a surprisingly high rate given that the GALEX catalog covers only a little more than half of the sky¹. GALEX DR5 provides precise farultraviolet (Fuv; 1344-1786 ˚A) and near-ultraviolet (Nuv; 1771-2831 ˚A) magnitudes with errors generally about 0.02-0.03 mag. We thus obtain a large collection of astrometric standards with good parallaxes and UV magnitudes which can be used to compute absolute ultraviolet magnitudes:

where ϖ is the parallax in mas. The uncertainty of absolute magnitudes is dominated by the error of Nuv magnitude for most of the stars, but distant objects (small parallaxes) can have the ratio ϖ/σ _ϖ close to unity, to the point that the observed parallax takes a negative value. To reduce the astrometric noise component in the subsequent analysis, the sample needs to be limited to the most reliable determinations with large ϖ/σ _ϖ or, which is almost equivalent in this case, with large parallaxes.

2. THE HERTZSPRUNG-RUSSELL DIAGRAM IN THE NEAR-ULTRAVIOLET

Figure 1 displays the “absolute Nuv magnitude versus Nuv−G color” (HR) diagram for 1403 stars selected with the logical and, or intersection, of the following criteria: ϖ/σ _ϖ > 5; ̟ϖ > 25 mas. Although observed BT and VT magnitudes are available for all Tycho-2 stars, as well as derived Johnson B and V magnitudes, I will use the more accurate broadband G magnitudes as observed by Gaia. For the sample under consideration, the distribution of formal errors of G magnitudes peaks at 0.0005 mag with a median of 0.0009 mag. This is much smaller than the uncertainty of Nuv magnitudes. The selection includes stars confidently within 40 pc of the Sun.

Fig. 1 Near-UV HR diagram of absolute Nuv magnitudes versus Nuv−G colors for nearby stars with parallaxes greater than 25 mas, matched with GALEX sources. The single dot far to the left represents the only single white dwarf in the sample, DN Dra. The line along the lower envelope of the main sequence shows the formal functional fit, see text. 

Most of the stars lie on a well-defined and narrow main sequence stretching between magnitudes 9 and 22 in absolute magnitude Mnuv and 5 - 12 in Nuv−G color. There is a rudimentary giant branch veering off to the right at the top of the main sequence, reflecting the scarcity of giants in the immediate solar neighborhood. The width of the main sequence is roughly 0.5 mag, likely to come from unresolved binaries. A regular MS-MS binary is shifted up and to the right of the main sequence because there is more additional flux in the G band than in Nuv. The largest deviation from the main sequence due to binarity is ≈ 0.75 mag for identical twin pairs. Interstellar reddening is not expected to have a significant presence in this diagram as there are no dense dust-molecular clouds within 40 pc.

Using a standard nonlinear fit algorithm (minimizing the residual RMS), this functional form is found for the lower envelope main sequence, represented with a solid curved line in Figure 1:

where x = (Nuv−G−8)/3.3. This curve accurately represents the bluest magnitudes and colors of normal field dwarfs without any signs of activity or reddening. The presence of a cosine term is justified by the much better results achieved: the standard deviation of post-fit residuals over a set of nodal points goes down from 0.21 to 0.08 mag, and the observed wiggles of the lower envelope are much more truthfully represented. A median main sequence functional expansion can be obtained from Equation 2 by replacing the constant term 15.339 with 15.150.

The inverse main sequence fit, i.e., [Nuv − G](Mnuv) may be handy if we want to estimate the amount of observed “UV-excess” for a known absolute magnitude:

where y = (Mnuv − 15.5)/6.5.

A few dozen stars lie to the left of the lower boundary curve with either their colors too blue or absolute magnitudes too faint. The latter is unlikely because of the slope of the main sequence - a deficit of Nuv flux would shift the point to the right of the main sequence. Hence, the stars below and to the left of the main sequence envelope have ultraviolet excess with respect to normal luminosities. This is confirmed by Figure 2 which shows a similar HR diagram for the same set of stars, but with Fuv magnitudes instead of Nuv. The main sequence is not well defined in the far-UV, but the stars with a large Nuv luminosity excess occupy a specific area of the diagram with absolute Fuv magnitudes greater than 6.7 and Fuv−G colors less than 11.6. This confirms that the ultraviolet excess for a fraction of nearest dwarfs is real and present in a wide range of wavelengths.

Fig. 2 Far-UV HR diagram of absolute Fuv magnitudes versus Fuv−G colors for nearby stars with parallaxes greater than 25 mas, matched with GALEX sources. The single encircled dot in the lower left corner represents the only single white dwarf in the sample, DN Dra. Stars of significant Nuv luminosity excess collected in Table 1 are also marked with circles. 

3. NEARBY STARS WITH NUV EXCESS

Table 1 lists nearby stars (distance less than 40 pc) found in the TGAS-GALEX sample with significant Nuv luminosity excess. The latter was defined as MNuv(obs)−MNuv(fit)> 0.7 mag. This is a conservative limit possibly leaving out a number of genuine sources of enhanced UV radiation, but it results in a more manageable sample of 50 stars which can be individually verified. The columns of the table include: (1) RA J2015 in degrees; (2) Dec J2015 in degrees; (3) HIP number when available; (4) Tycho-2 identification when HIP number is not available; (5) parallax in mas; (6) standard error of parallax in mas; (7) G magnitude; (8) Fuv magnitude, if available; (9) Nuv magnitude; (10) formal error of Fuv magnitude, if available; (11) formal error of Nuv magnitude. Columns 1 through 7 are copied from TGAS, while Columns 8 - 11 are copied from GALEX.

The single point far to the left in Figures 1 and 2 represents the well-known white dwarf DN Dra = GJ 1206 of spectral type DA4.0 (e.g., ^{Fontaine et al. 1992}). It is very luminous in the near-UV with an absolute magnitude Mnuv= 7.95 mag. The absence of other bright white dwarfs within 40 pc of the Sun in our selection is probably explained by selection effects in the Hipparcos, Tycho-2, and TGAS catalogs². Other excess stars have much redder Nuv−G colors and cannot be isolated white dwarfs. An extensive literature and astronomical database search with VizieR and Simbad reveals that the sample includes predominantly dwarfs of late K to early M spectral types. Some of these stars are included in the study of the near-UV luminosity function of early M-type dwarfs by ^{Ansdell et al. (2015)}, where the authors used Nuv fluxes relative to visual and near-infrared fluxes rather than absolute luminosities, which leads to a larger sample. ^{Ansdell et al. (2015)} find that up to 1/6 of all such M dwarfs show elevated levels of near-UV radiation, which may be inconsistent with a constant star-formation rate and commonly used age-activity relations. Here we find a much lower rate of dwarfs with excess Nuv lumnosities in absolute units (∼ 3.6%). It is possible that a relative-flux selection is biased toward more active M dwarfs from a larger volume of space.

3.1. Too Many Young Stars?

All of our late-type dwarfs satisfy the rather generous selection criteria for young stars of ^{Rodriguez et al. (2013)}, their Fig. 1. Can they all be young? Assuming a constant rate of star formation over the 13 Gyr history of the Galaxy, the rate of overluminous dwarfs corresponds to a threshold age of 460 Myr. Hence, the existence of such dwarfs in the solar neighborhood can be explained if stars younger than the Hyades can retain the observed Nuv excess due to a high level of magnetic activity fueled by fast rotation. There are no star forming regions, OB associations, or young open clusters within the close solar neighborhood. However, some of the stars listed in Table 1 have been proposed as members of sparse young moving groups (YMG). Some interesting examples are:

• TYC 5899-26-1, an M3.3 dwarf, was assigned by ^{Shkolnik et al. (2012)} to the AB Doradus YMG with an estimated age of 30-50 Myr (^{Makarov 2007}).

• TYC 5832-666-1, a rotationally variable M0 dwarf, was assigned by L´epine & Simon (2009) to the β Pic YMG with an estimated age of 20-30 Myr.

• HIP 84794 = GJ 669A, a flaring M3.5 dwarf, was assigned by ^{Shkolnik et al. (2012)} to the Hyades MG with an estimated age of 600 Myr³.

• HIP 112312 = WW PsA, an M1 dwarf, was assigned by ^{Shkolnik et al. (2012)} to the β Pic MG, but it is also a rotationally variable binary of the BY Dra type.

• HIP 102409 = GJ 803 = AU Mic, a famous M1e young dwarf with a resolved debris disk, considered to be a member of the β Pic MG, but also an active binary of the BY Dra type.

• TYC 6351-286-1 = HD 201919, a rotationally variable K6Ve dwarf suggested by ^{Elliott et al. (2016)} as a member of the AB Doradus YMG.

^{Shkolnik & Barman (2014)} conclude that the median UV flux of early M stars remains at “saturated” levels for a few hundred Myr, and the decline in activity after ≈ 300 Myr follows a time⁻¹ dependence, but their analysis is based on rather rough distance estimates and the relative FUV/FJ flux ratio. Often, the proposed membership of stars to the nearest moving groups is uncertain and suffers from considerable rates of interlopers. The earlier attempts at identifying such groups were based on proper motion and X-ray count rate data following the successful completion of the Rosat and Tycho-2 missions (^{Makarov & Urban 2000}). But the census of nearby most luminous stars in X-rays shows that this criterion nets more active binaries of the RS CVn and BY Dra type than very young objects (^{Makarov 2003}). Even though the majority of objects in Table 1 are associated with Rosat-detected X-ray sources, this does not guarantee their young age. Figure 3 presents an attempt to verify that the over-luminous dwarfs can be younger than the Hyades. Only 9 known Pleiades members seem to be present in the TGAS-GALEX sample, marked with open circles. These stars are solar-type or earlier, and they conform to the main sequence fit quite well. Unfortunately, small-mass dwarfs are missing, perhaps because they are too faint. The filled circles represent the proposed members of the nearer and possibly younger Tuc Hor MG (estimated age 27 Myr) from ^{Makarov (2007)}; ^{Kraus et al. (2014)}. They allow us to probe later spectral types down to the early K. These candidate young stars start to deviate from the main sequence at Mnuv≈14-15 mag. This may be interpreted as a “turn-on” point of very young stars, which is likely age-dependent. The absence of late-type dwarfs thwarts verification of this result. The preliminary conclusion is that stars younger than the Pleiades (≲100 Myr) that wandered by chance into the close solar neighborhood may be significantly over-luminous in the UV compared to older inactive field stars, but their number should be much smaller than what we find on the HR diagram.

Fig. 3 Near-UV HR diagram of absolute Nuv magnitudes versus Nuv−G colors for known members of the Pleiades cluster (open circles) and Tuc Hor MG members (filled disks) found in the TGAS-GALEX cross-match. The diamond in the lower left corner shows the data for the nearby white dwarf van Maanen 2. The solid line is the main sequence fit for nearby stars (Equation 2). The dashed line shows the computed loci of old inactive dwarfs in unresolved binaries with white dwarfs identical to van Maanen 2. 

4. A WIDER SELECTION OF UV-LUMINOUS STARS

A broader search for genuine hot stars in TGAS can be made if we drop the small distance criterion and consider the entire population with statistically precise parallaxes and matching GALEX sources. Table 2 lists 40 stars found with the following criteria: ϖ/σ _ϖ > 5, and Nuv−G < 2 mag. The format of this table is the same as Table 1. Besides the previously found DN Dra, this selection includes one additional well-known WD of DA0.8 type, HIP 12031=FS Cet. With G = 12.177 mag, Nuv=12.371 mag, Mnuv= 7.95 mag, this star marks the top of the WD cooling sequence in the HR diagram. Between FS Cet and van Maanen 2, the range of absolute Nuv magnitudes of white dwarfs is ≈ [8,20] mag, and this should make the Nuv HR diagram a suitable proxy for the spectroscopic determination of type.

Most of the objects in Table 2 are relatively poorly studied stars that have remained under the radar of observers. It is only now with the combination of precise GALEX photometry and Gaia parallaxes that we begin to see them as very unusual objects. Several stars, on the contrary, have been studied in more detail, including:

The appearance of known novae and spectroscopic binaries with hot subdwarfs implies that more hidden WD and sdOB can be discovered among the relatively nearby objects listed in Table 2. Followup spectroscopic and photometric observations are perhaps the best way to find the nature of their excessive UV luminosity.

5. CONCLUSIONS

This study of nearby astrometric standards from the Gaia DR1 shows that stellar youth is only one of the reasons for field stars to have excess Nuv luminosities, and perhaps, not the main one. Dynamical and possibly magnetic interaction of low-mass dwarfs in close binaries is capable of supporting fast rotation rates and enhanced levels of X-ray and UV radiation for durations comparable to the main-sequence lifetimes. This is confirmed, for example, by in-depth investigations of stellar rotation rates in nearby open clusters. ^{Douglas et al. (2016)} find that almost all single members of the Hyades (age ≈ 650 Myr) with masses above 0.3M _⊙ are slow rotators, while most of the spectroscopic binaries in this mass range are fast rotators. Many of the nearby stars listed in Table 1 with Nuv−G colors bluer than the main sequence are expected to be old binary systems of BY Dra type. In a wider sample of stars with extreme UV colors listed in Table 2, the presence of binaries with white dwarf and sdOB hot subdwarf components is conspicious, but many others remain hidden.

It is reasonable to expect that metal-poor Population II stars should also show mild Nuv excess compared with disk dwarfs. The weak absorption lines of metals in the near-UV region provide additional flux at short wavelengths. Since the fraction of Population II stars in the solar neighborhood is low, we expect few, if any, such objects to be present in our analysis. Using tangential velocities (computed from Gaia proper motions and parallaxes) as a proxy for population type, a search for high-velocity stars within 40 pc of the Sun resulted in 70 objects (out of 1403) with v _tan > 70 km s⁻¹. These fast moving stars comply with the main sequence quite well (not shown in this paper for brevity) with the exception of a few objects deviating to the giant domain and possibly three stars with mild Nuv excess, all with Mnuv around 15 mag. Only one of the three objects satisfies the strict selection criteria adopted here, namely, the previously discussed spectroscopic binary HIP 45731, but there is no evidence of metal deficiency in the literature.

It is also found that most of the field stars in the immediate solar neighborhood (distance less than 40pc) follow a well-defined and narrow main sequence on the “absolute Nuv magnitude versus Nuv−G color” HR diagram constructed with Gaia parallaxes and GALEX and Gaia photometry. This confirms the high quality of GALEX and Gaia photometric data and makes such a diagram a valuable method to detect more stars with unusual UV radiation properties.

The author is grateful to J. Subasavage and J. Munn for useful discussions of the topic. This work has made use of data from the European Space Agency (ESA) mission Gaia (http://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, http://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This research has made use of the VizieR catalogue access tool, CDS, Strasbourg, France. The original description of the VizieR service was published in A&AS 143, 23.